PET/CT metabolic measures predict early breast cancer recurrence

Preoperative [¹⁸F]FDG PET/CT measures of tumor burden and glucose metabolism were associated with recurrence or metastasis within 24 months in a retrospective study of 106 women with early-stage breast cancer. Patients whose disease recurred or spread had greater metabolic tumor volume and total lesion glycolysis in the primary breast tumor, higher maximum standardized uptake value in axillary lymph nodes, and larger whole-body metabolic volumes than those who remained recurrence-free. A combined analysis of the imaging measures produced an area under the receiver operating characteristic curve of 0.805, indicating moderately good discrimination in this single-center cohort.

The study included women treated surgically at Beijing Shijitan Hospital between January 2021 and December 2023. All had clinical stage I or II breast cancer, defined as T1–2, N0–1, M0, and underwent PET/CT within one week before surgery. Patients who received neoadjuvant chemotherapy, radiotherapy or targeted therapy were excluded, as were those with previous malignancy, severe underlying disease, active infection or imaging that could not be interpreted reliably.

Follow-up was conducted by telephone and outpatient visits for 24 months. Of the 106 participants, 57 had no recurrence or metastasis, while 49 experienced a confirmed event, including lymph-node metastasis in 19 patients, chest-wall recurrence in six, pleural metastasis in three, bone metastasis in 15 and lung metastasis in six. Imaging was performed with a Discovery 710 PET/CT scanner.

Patients fasted for six to eight hours, received an intravenous dose of 4.44–5.18 MBq/kg of [¹⁸F]FDG and rested for 50–60 minutes before scanning. A low-dose CT was acquired from head to thigh, followed by three-dimensional PET acquisition. Images were reconstructed with an ordered-subsets expectation-maximization algorithm using two iterations, eight subsets, a 128×128 matrix and a 5-mm Gaussian filter.

Quantitative analysis was performed with a three-dimensional volume viewer to derive standardized uptake values, metabolic tumor volume and total lesion glycolysis for the primary tumor, axillary lymph nodes and whole-body region. Two experienced nuclear-medicine physicians assessed the ipsilateral axillary nodal basin. When no FDG-avid node could be identified, investigators placed a standardized circular region of interest over that basin, recorded its maximum standardized uptake value and assigned nodal metabolic tumor volume and total lesion glycolysis values of zero.

The recurrence or metastasis group had a primary-tumor metabolic tumor volume of 21.95 ± 6.72 cm³, compared with 18.27 ± 5.33 cm³ in the non-recurrence group, a statistically significant difference with P = 0.002. Primary-tumor total lesion glycolysis was also higher, at 100.47 ± 32.18 g versus 86.84 ± 20.25 g, with P = 0.012. Maximum standardized uptake value in axillary lymph nodes averaged 3.74 ± 1.18 in the recurrence group and 3.18 ± 0.95 in the non-recurrence group, with P = 0.008.

Whole-body metabolic tumor volume and total lesion glycolysis were similarly elevated among patients who recurred: 30.06 ± 8.14 cm³ versus 26.32 ± 7.83 cm³, P = 0.018, and 119.29 ± 30.67 g versus 103.58 ± 26.42 g, P = 0.006. Not every PET/CT measure distinguished the groups. Primary-tumor maximum and mean standardized uptake values, axillary-node mean uptake, axillary-node metabolic volume and axillary-node total lesion glycolysis did not differ significantly.

Whole-body maximum and mean standardized uptake values also failed to reach statistical significance. The results therefore pointed more strongly to measures that combine FDG uptake with the amount of tissue involved than to peak uptake alone. The study also found differences in surgical pathology: lymphovascular invasion was present in 15 of 49 patients in the recurrence group, compared with eight of 57 in the non-recurrence group.

Higher Ki-67 expression, lower estrogen- and progesterone-receptor positivity, greater HER2 positivity and a higher frequency of triple-negative disease were also observed among patients with adverse outcomes. Because metabolic tumor volume and total lesion glycolysis were strongly correlated, including all five PET/CT measures in one model would have created severe multicollinearity. The investigators therefore fitted five separate multivariable logistic-regression models, each containing one imaging measure plus lymphovascular invasion and pathological nodal stage.

Primary-tumor metabolic tumor volume remained associated with recurrence after adjustment, with an odds ratio of 1.150 per additional cm³, a 95% confidence interval of 1.052–1.257 and P = 0.002. A 10 cm³ difference corresponded to an odds ratio of 4.046, with a 95% confidence interval of 1.657–9.875. Primary-tumor total lesion glycolysis had an odds ratio of 1.035 per gram; expressed as a 30-g difference, the odds ratio was 2.806, with a 95% confidence interval of 1.492–5.276 and P = 0.002.

Axillary-node maximum standardized uptake value also retained an association in its adjusted model, with an odds ratio of 1.700 per unit, a 95% confidence interval of 1.050–2.752 and P = 0.031. Whole-body total lesion glycolysis was associated with recurrence, with an odds ratio of 1.028 per gram, a 95% confidence interval of 1.008–1.049 and P = 0.006; a 30-g difference corresponded to an odds ratio of 2.290, with a 95% confidence interval of 1.257–4.172. Lymphovascular invasion was independently associated with recurrence in most models, whereas pathological N stage was not independently associated in any of them.

The paper also reports whole-body metabolic tumor volume as an independent factor, although the reported P value and confidence interval for that estimate are not fully consistent, so its strength requires clarification in validation work. Receiver operating characteristic analysis indicated that the individual imaging measures had only limited discrimination on their own. Primary-tumor metabolic tumor volume produced an area under the curve of 0.672, with an optimal cutoff of 23.755 cm³, 40.8% sensitivity and 87.7% specificity.

Primary-tumor total lesion glycolysis had an area under the curve of 0.612, while axillary-node maximum standardized uptake value had an area under the curve of 0.631. Whole-body metabolic tumor volume and total lesion glycolysis yielded areas under the curve of 0.638 and 0.650, respectively. The combined measure reached an area under the curve of 0.805, showing that the imaging variables collectively separated the two outcome groups better than the individual parameters, although the study does not establish performance in an independent population.

Exploratory subgroup analyses suggested that the associations persisted among the 85 patients with invasive ductal carcinoma and across surgery types. Four imaging parameters remained associated with recurrence among 44 patients who underwent breast-conserving surgery, and similar associations appeared among 62 patients who received mastectomy. Interaction tests were not significant, but only 21 patients had histologic types other than invasive ductal carcinoma, too few for stable multivariable estimation.

The results were not restricted to a particular surgical approach, yet the retrospective design, single institution and modest sample size prevent the subgroup observations from establishing equivalence between treatments. Detailed information about postoperative adjuvant regimens was also not systematically collected, leaving potential treatment-related confounding unresolved. These findings support PET/CT-derived metabolic measures as possible research markers of early recurrence risk, not as a validated clinical test.

The cohort was restricted to women with stage I–II disease who proceeded directly to surgery, and outcomes were assessed over only 24 months, so the results do not describe later recurrence or the full spectrum of early-stage breast cancer. Model development and evaluation occurred in the same cohort, with no external validation. Measurement also varied for axillary nodes: when no FDG-avid node was visible, the investigators used a standardized region of interest over the nodal basin and recorded metabolic volume and glycolysis as zero, a procedure that may introduce variability.

Prospective studies with larger and more diverse populations, longer follow-up and standardized acquisition and analysis are needed before these measures can guide surveillance or treatment.

Subject of Research: Preoperative FDG PET/CT metabolic parameters as predictors of early postoperative recurrence and metastasis in early-stage breast cancer

Article Title: Predictive value of preoperative [18F]FDG PET/CT imaging parameters for postoperative recurrence and metastasis in breast cancer patients

Article References: Lin, N., Ren, C., & Wen, Z. (2026). Predictive value of preoperative [18F]FDG PET/CT imaging parameters for postoperative recurrence and metastasis in breast cancer patients. Frontiers in Oncology, 16, Article 1946005. https://doi.org/10.3389/fonc.2026.1946005

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1946005

Keywords: breast cancer, FDG PET/CT, tumor recurrence, cancer metastasis, metabolic tumor volume, total lesion glycolysis, lymphovascular invasion, prognostic imaging, early-stage cancer, Predictive, value, preoperative